Human Reproduction Vol.22, No.5 pp. 1310–1319, 2007 doi:10.

1093/humrep/dem003
Advance Access publication February 20, 2007

The effectiveness of a fertility awareness based method to

avoid pregnancy in relation to a couple’s sexual behaviour
during the fertile time: a prospective longitudinal study

P.Frank-Herrmann1,8†, J.Heil2,†, C.Gnoth3, E.Toledo4, S.Baur5, C.Pyper6, E.Jenetzky7,

T.Strowitzki1 and G.Freundl5
1
Department of Gynaecological Endocrinology and Fertility Disorders, University of Heidelberg, Voßstrasse 9, 69115 Heidelberg,
Germany, 2Department of Obstetrics and Gynaecology, University of Heidelberg, Voßstrasse 9, 69115 Heidelberg, Germany,
3
Department of Obstetrics and Gynaecology, University of Cologne and Centre for Family Planning, Endocrinology and
Reproductive Medicine, Rheydter Straße 143, 41515 Grevenbroich, Germany, 4Department of Preventive Medicine and Quality
Management. Hospital Virgen del Camino, C/Irunlarrea 4, E-31008 Pamplona, Spain, 5Section Natural Fertility, German Society for
Gynaecological Endocrinology, Voßstrasse 9, 69115 Heidelberg, Germany, 6National Perinatal Epidemiology Unit, Department
of Public Health, University of Oxford, Old Road Campus, Oxford, OX3 7LF, Great Britain and 7Institute of Medical Biometry and

Downloaded from http://humrep.oxfordjournals.org/ by guest on June 17, 2013

Informatics, University of Heidelberg, Im Neuenheimer Feld 305, 69120 Heidelberg, Germany
8
To whom correspondence should be addressed at: Department of Gynaecological Endocrinology and Fertility disorders, University of
Heidelberg, Voßstrasse 9, 69115 Heidelberg, Germany. E-mail: petra.frank-herrmann@med.uni-heidelberg.de

BACKGROUND: The efficacy of fertility awareness based (FAB) methods of family planning is critically reviewed.
The objective was to investigate the efficacy and the acceptability of the symptothermal method (STM), an FAB
method that uses two indicators of fertility, temperature and cervical secretions observation. This paper will rec-
ommend a more suitable approach to measure the efficacy. METHODS: Since 1985, an ongoing prospective observa-
tional longitudinal cohort study has been conducted in Germany. Women are asked to submit their menstrual cycle
charts that record daily basal body temperature, cervical secretion observations and sexual behaviour. A cohort of 900
women contributed 17 638 cycles that met the inclusion criteria for the effectiveness study. The overall rates of unin-
tended pregnancies and dropout rates have been estimated with survival curves according to the Kaplan–Meier
method. In order to estimate the true method effectiveness, the pregnancy rates have been calculated in relation to
sexual behaviour using the ‘perfect/imperfect-use’ model of Trussell and Grummer-Strawn. RESULTS: After
13 cycles, 1.8 per 100 women of the cohort experienced an unintended pregnancy; 9.2 per 100 women dropped out
because of dissatisfaction with the method; the pregnancy rate was 0.6 per 100 women and per 13 cycles when
there was no unprotected intercourse in the fertile time. CONCLUSIONS: The STM is a highly effective family
planning method, provided the appropriate guidelines are consistently adhered to.

Key words: contraception/efficacy/fertility awareness based methods/natural family planning/symptothermal method

Introduction and second the modification of sexual behaviour either to

Background
Fertility awareness based (FAB) methods is a term that
includes all family planning methods that are based on the
identification of the fertile time. They are based on the
woman’s observation of physiological signs of the fertile and
infertile phases of the menstrual cycle. This knowledge can
be used to plan or avoid pregnancy. FAB methods depend on
two key variables: first the accurate identification of the
fertile days of a woman’s menstrual cycle (the fertile time)
†
The first two authors contributed equally to this work.
plan a pregnancy or to use this knowledge to avoid pregnancy.
When couples use FAB methods of family planning to avoid
pregnancy, they practise different sexual behaviour during
the fertile time. When FAB methods involve sexual abstinence
during the fertile time, this method is called natural family
planning (NFP). When FAB methods involve occasionally
using a barrier method during the fertile time, the method is
called FAB method with barriers. It must also be recognized
that although many couples state they are practising a FAB
method, sometimes they do not adhere to the guidelines and
unprotected intercourse or other kinds of genital contact

1310 # The Author 2007. Published by Oxford University Press on behalf of the European Society of Human Reproduction and Embryology.
All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org

occur during the fertile time. The efficacy of FAB methods to
avoid pregnancy has been critically reviewed by several
authors (Fehring et al., 1994; Frank-Herrmann et al., 1991;
De Leizaola-Cordonnier, 1995; Barbato and Bertolotti, 1988;
Hilgers and Stanford, 1998; Howard and Stanford, 1999;
Kambic, 1999; The European Natural Family Planning Study
Groups, 1999; Grimes et al., 2004). Several issues have been
identified when attempting to compare the different FAB
methods.
The first is that most FAB methods have evolved concur-
rently over the last 40 years in different countries; each has
been lead by pioneers who have developed guidelines for
their respective groups. This has resulted in many cases in a
lack of evidence-based guidelines being developed and
subsequently modified to conform to best scientific evidence.
The second issue is that efficacy rates may vary because they
are derived from studies done with volunteers and researchers
from different cultural backgrounds where motivation to avoid
pregnancy and rigour of research methods may vary (World
Health Organization, 1981a,b; Gomes and Congdon, 1988;
Xu et al., 1994; Indian Council of Medical Research Task
Force on Natural Family Planning 1996).
Third, the methods currently used to calculate the method
effectiveness are questionable. Many investigators have recog-
nized the importance of distinguishing between pregnancies
attributable to user failure and method failure (method effec-
tiveness). The standard procedure up to now was to compute
separate method and use-effectiveness rates (pregnancies
divided by exposure). In this procedure, all exposure from
perfect and imperfect use is included in denominator of both
method and user failure rates. The common misinterpretation
is that the resulting method effectiveness rate yields infor-
mation about the inherent efficacy of the method. Inherent
method efficacy can be measured only when the numerator
(method failures) is assessed in relation to the proper risk set,
i.e. the exposure only when the method is used perfectly. For
this reason, method effectiveness rates computed by the stan-
dard procedure are biased downwards to an unknown extent
(Trussell and Grummer-Strawn, 1990).
Figure 1. Determination of the fertile time according to the guidelines of the symptothermal (STM) method.
Effectiveness of a fertility awareness-based method
This problem is further confounded by the different ways an
unintended pregnancy is classified. Some prospective studies
ensure the couple’s intention to avoid a pregnancy is recorded
at the beginning of each menstrual cycle. Other studies are ret-
rospective and only question the couple’s intention after sexual
intercourse has been recorded during the fertile time.
The fourth issue is that some new FAB methods are simpli-
fied methods that are often used in developing countries and
very relevant for settings where cost of teaching is an issue and
where continuation has a higher priority than efficacy (Thapa
et al., 1990; Jennings and Sinai, 2001; Arevalo et al., 2004).
To be able to make an informed choice when selecting a
family planning method, couples need to know the efficacy
of a method when used consistently and imperfectly. Trussell
and Grummer-Strawn are critical about how efficacy has
been calculated in previous studies. They argue that previously
published rates of method and user failure for all contraceptive
methods suffer from methodological errors and are biased
downwards. Trussell and Grummer-Strawn (1990, 1991) rec-
Downloaded from http://humrep.oxfordjournals.org/ by guest on June 17, 2013
ommend a new model of calculating perfect and imperfect
use pregnancy rates which up to now has rarely been applied.
This is likely to be due to the fact that this approach requires
documentation of all sexual behaviour during each cycle.
Description of the method
This paper describes a cohort of couples who used a method
that consisted of recording the cervical secretion pattern,
changes of basal body temperature and the application of a cal-
culation rule. It is called the symptothermal method (STM) of
NFP. The beginning and the end of the fertile time are ident-
ified by two parameters in order to have double-check system.
The following two guidelines are given to each couple to
identify the first fertile day – both guidelines are applied and
the first fertile day is the earliest day identified (Figure 1):
(i) Change of cervical secretion: first appearance of cervi-
cal secretion.
(ii) Calculation guideline: the first fertile day is the sixth
day of the cycle (In NFP methodology, this is called
the ‘five days rule’ which states that the first five
1311
P.Frank-Herrmann et al.
days of the cycle are infertile days.) or after a woman
has completed 12 cycles of use this guideline is
replaced by a calculation that takes the earliest temp-
erature rise in the last 12 cycles and subtracts seven
days to identify the first fertile day (In NFP method-
ology, this is called the ‘minus eight rule’: earliest
temperature rise in the last 12 cycles minus eight
days to identify the last infertile day.)
The following two guidelines are given to each couple to ident-
ify the last fertile day—both guidelines are applied and the last
fertile day is the latest day identified (Figure 1):
(i) The evening of the third day after the cervical secretion
peak day (The cervical secretion peak day is only
recognized on the day following peak, when the
secretions have become sticky again.)
(ii) The evening of third higher temperature reading, all
three higher than the previous six readings, the last
one 0.28C higher than the previous six.
These evidence-based guidelines have been developed follow-
ing extensive research that has been carried out over the last 20
years. They have been adopted widely by FAB groups who
teach the STM method (Raith et al., 1999). Detailed guidelines
of the STM methodology are described elsewhere (Arbeits-
gruppe NFP, 2006).
Objectives
Our first objective was to analyse the overall use effectiveness
of the STM method and to determine whether the effectiveness
was different for different types of sexual behaviour during the
fertile time according to the ‘perfect/imperfect-use model’
(Trussell and Grummer-Strawn, 1990). In particular, we
specifically wanted to know if those women who only used
the STM without combining with a barrier method during the
fertile time had fewer unintended pregnancies, regarding the
use-effectiveness rates, than those women who occasionally
used barrier methods during the fertile time.
Our second objective was to investigate the acceptability of
the STM method. We therefore reviewed those couples whose
reasons for discontinuing the STM were due to dissatisfaction
with the method or due to difficulties with using the STM.
Materials and methods
The study was conducted by the German NFP study centre in
accordance with the ethical principles of the Declaration of Helsinki.
The study centre is an integral part of the German Society of Gynae-
cological Endocrinology and Reproductive Medicine. Its main aim is
to undertake research studies in the field of NFP and to evaluate NFP
services. For the last 20 years, the study centre has coordinated this
prospective, observational longitudinal cohort study. Between 1985
and 2005, the study enrolled 1599 women using the STM in different
situations and collected data from 35 996 menstrual cycles. This data
set has already addressed several questions (Gnoth et al., 1999, 2002,
2003). From this data set, a cohort of 900 women with 17 638 cycles
met the effectiveness study selection criteria.
Exclusion criteria
All couples with a reason of potential sub- and infertility, or not being
exposed to risk of conception or trying for pregnancy with the help of
1312
Table I. Exclusion criteria and number of participants excluded
Category Number Percentage
Participants in the whole database 1599
Participants in the effectiveness study 900
Excluded 699 100
Participants excluded by reason
Pregnancy achieving at study entry 356 50.9
Post-pill 125 17.9
Irregular cycles (.35 days) 74 10.6
Post-partum/breastfeeding 59 8.5
Experienced users 33 4.7
Premenopausal women over 45 years 27 3.9
No sexual partner 12 1.7
Young women under 19 years 8 1.1
Post-abortum 5 0.7
NFP were excluded. The reasons for excluding the women and the
number of women excluded are given in Table I.
Inclusion criteria
Downloaded from http://humrep.oxfordjournals.org/ by guest on June 17, 2013
The inclusion criteria are given in Table II in accordance with the
recommendations from Tietze and Lewit (1974), Trussell and Kost
(1987) and Potter (1996). A subcohort of 900 women with 17 638
cycles was selected out of the whole database.
The women had to state that they intended to avoid pregnancy and if
they changed their intention they had to be willing to let the
investigator know at once in order to have a clear definition of
unintended pregnancy (discussed later). The women had to be
willing to record all sexual behaviour, especially the occasional use
of barrier methods to ensure clarity about the definitions for perfect/
imperfect-use (discussed later). The couples must not use any contra-
ceptive methods other than barrier methods. The study specifically
only included those couples who were starting to use the STM
method and commencing the first STM cycle. To ensure the partici-
pants were likely to have normal fertility, the women had to be
between 19 years at entry into the study and less than 46 years at
the end of the study. To ensure the women could become pregnant
(i.e. were likely to have fertile ovulations), the average cycle length
of the women had to be between 22 and 35 days (20% of the cycles
of each study participant was allowed to deviate outside this range).
Only those women with no previous history of infertility were
included. There was no requirement for proven fertility in terms of
the women already having a history of being pregnant, in order not
to exclude the younger, potentially more fertile and sexually active
women. In a previous data analysis, we found no significant difference
in unintended pregnancy rate between those with and without proven
Table II. Inclusion criteria for the effectiveness study
Inclusion criteria
Age 19– 45 years
Normal cycle lengths between 22 and 35 days (20% of cycle lengths could
be outside this range)
Willing to record family planning intention at the start of each cycle
Willing to record sexual behaviour, including sexual intercourse, genital
contact, withdrawal, occasional barrier use
Agreement not to use any other forms of contraception
No known history of subfertility or infertility
An established luteal phase of at least 10 days hyperthermic phase and at
least 3 months following breastfeeding, oral contraception, post-partum,
post-abortum
Willing to participate in the study for 12 months

fertility (Frank-Herrmann et al., 1991). Women who had either
delivered a child or breastfed or used oral contraceptives were only
included after 3 months of an established luteal phase, diagnosed by
an elevated temperature phase for at least 10 days. All the women
were asked to agree to participate for at least 12 cycles.
Teaching the STM
All the women who participated in the study were taught the STM by
accredited teachers from the ‘Arbeitsgruppe NFP’ which was founded
in 1981 with the aim of promoting NFP in Germany. In collaboration
with the German NFP study centre, the training and the teaching meth-
odology was standardized and adhered to strict guidelines. There were
comprehensive teaching materials that accompanied the personal
small group teaching sessions (Arbeitsgruppe NFP, 2006).
Recruitment of study participants
The participants were volunteers who had self-selected to join the
study following given standardized information about the study by
their STM teachers; all women gave their informed consent. Standar-
dized admission questionnaires were used to collect relevant data con-
cerning age, parity, family planning history and socio-demographic
background. All women were asked to send their cycle charts, after
each cycle was completed, that recorded basal body temperature,
quality of cervical secretions, cycle length, family planning intention
and sexual behaviour directly to the study centre. Those women,
whose cycle charts did not reach the study centre, were contacted on
three separate occasions by the study centre via the woman’s personal
teacher. If a woman did not respond to any of the requests for infor-
mation she was then classified as lost to follow-up.
Discontinuation
We were specifically investigating the acceptability of the STM.
Therefore, the most important reasons given for discontinuation
were those due to dissatisfaction with the STM, and/or change to
other family planning method.
Table III lists all the reasons for discontinuation.
Data collection
The software used to handle the data was a Microsoft Accessw rela-
tional database system called NFP DAT 1.0; it is described in a pre-
vious paper (Gnoth et al., 1999). It has an automated analysis
system that follows-up the participant every 3 months.
Definition of the pregnancy
The definition of pregnancy was an elevated temperature of longer
than 18 days and clinical pregnancy test confirmed by the researcher.
Table III. Reasons for discontinuation during the effectiveness study
Reasons for discontinuation
Desire to get pregnant
Unintended pregnancy
Discontinuation because the couple was dissatisfied with the method:
discomfort with the method, problems with observing the indicators of
fertility, feeling of insecurity, finding the fertile time too long, finding it
difficult to abstain during this time
Change to other family planning method
The woman does not want to be part of the study any longer; however, she
will continue practising the symptothesmal (STM) method
Medical or surgical reasons (e.g. Hysterectomy)
Separation from partner
Effectiveness of a fertility awareness-based method
All the pregnancy charts were reviewed and confirmed by the scienti-
fic committee of the NFP study centre.
Definition of the unintended pregnancy
Pregnancies were classified as intended or unintended on the basis of
the statements made by the women before conception. At the end of
each menstrual cycle, the woman was asked to state if she was plan-
ning to become pregnant the following cycle. This was documented
in the completed cycle chart. If she forgot to answer this question,
and if a pregnancy occurred in the next cycle, it was always classified
as an unintended pregnancy. If charts did not reach the study centre in
time, the last indicated family planning intention held at the study
centre was used to classify a pregnancy as intended or unintended.
Data analyses
Statistical analyses were carried out using the SASw package, version
8. We used the non-parametric model of Kaplan–Meier, the survival
curve or actuarial curve, to estimate the rates of unintended pregnancy,
the drop out due to dissatisfaction and the women lost to follow-up
(Kaplan and Meier, 1958; Matthews and Farewell, 1996). We
defined the ‘survival’ of a woman as the duration in the study until
Downloaded from http://humrep.oxfordjournals.org/ by guest on June 17, 2013
she dropped out for the target event (¼ unintended pregnancy).
Other dropouts are censored. The time unit was the menstrual cycle,
therefore, the estimated rates correspond to the life table approach
of Tietze and Potter, often used in earlier family planning studies
(Potter, 1966; Tietze and Lewit, 1974). In contrast to the Pearl-index,
the actuarial curves according to Kaplan–Meier represent a time-
related hazard estimation. The results at observation cycle 13 can
roughly be compared to the Pearl-index (¼ number of unintended
pregnancies per 100 women years, defining 13 cycles to be one
woman year). To compare the actuarial curves of different parameters,
a logrank test was performed. Chi-squared-test was used for categori-
cal data.
The Kaplan–Meier approach was used to calculate the overall
effectiveness rates. Pregnancies due to both method and user failure
were included.
In order to calculate the method effectiveness, we used a modified
model of the ‘perfect/imperfect-use’ approach (discussed earlier,
Trussell and Grummer-Strawn, 1990): pregnancy rates were calcu-
lated according to sexual behaviour: all unintended pregnancies that
occurred during a defined mode of sexual behaviour were related to
all cycles of the corresponding type of sexual behaviour,
i. e. unintended pregnancies that occurred in cycles with protected
intercourse during the fertile time were related to only those cycles
with protected intercourse in the fertile time. According to this
approach, we defined the following categories of sexual behaviour
in the fertile time: only abstinence, only protected intercourse,
protected and unprotected intercourse, only unprotected intercourse,
coitus interruptus or genital contact.
Results
To evaluate the overall use-effectiveness as well as the
method-related discontinuation and lost to follow-up, we
studied the following groups:
(i) Out of the whole cohort of 900 women and 17 638
cycles, we calculated the overall use-effectiveness as
well as the method-related discontinuation, the lost to
follow-up and the overall duration of study
participation.
(ii) This whole cohort was divided into two subgroups: 322
women used only the STM (‘STM only’-group) and
1313
P.Frank-Herrmann et al.

509 women used the STM with occasional use of bar-
riers in the fertile time (‘STM mix’-group). The ‘STM
mix’-group used a barrier method in 53% of their
cycles. Life table pregnancy rates have been analysed
separately for these two groups. Sixty-nine women
did not document their sexual behaviour and were
therefore excluded from that analysis.
To analyse the pregnancies according to the modified ‘perfect/
imperfect-use’ model, we formed different categories as
described earlier.
Client profile
The socio-demographic characteristics of the study population
are shown in Table IV More than 60% of the women were
between 19 and 29 years old. Nearly two-thirds of the
women had a medium educational level (German baccalaureate
or equivalent without a university degree), 52% were nulligra-
vidas; about 20% had reached their desired family size and
nearly 60% of the women wanted a further child in the future
a univariate point of view, we calculated slightly different
rates of 1.62 (1/2 0.89) for the STM only-group versus
2.02 (1/2 0.72) for the ‘STM mix’ group at 13 cycles –
which corresponds approximately to 1 year. At 24 cycles the
differences were inversed. These differences were not found
to be statistically significant at any time (Logrank test:
x2 , 0.31, hence P . 0.60).
Figure 2 illustrates the overlapping standard errors.
Pregnancy rates in relation to sexual behaviour
in the fertile time
In order to accurately estimate the true method effectiveness
according to the ‘perfect/imperfect-use’ approach, every type
of sexual behaviour had to be documented. Charting of
sexual behaviour occurred in 85% of the cycles; analyses of
these cycles showed that in more than a third the STM was
used with abstinence during the fertile time, which reflects
the ‘perfect-use’ scenario and true method effectiveness. For
perfect use, the unintended pregnancy rate was 0.43 per 100

Downloaded from http://humrep.oxfordjournals.org/ by guest on June 17, 2013

but not during the year of the study. women and 13 cycles (Table VII).
In contrast, the rate of unintended pregnancies in cycles with
Overall unintended pregnancy rates (use-effectiveness) unprotected intercourse during the fertile time the unintended
For the whole cohort, we calculated an unintended pregnancy pregnancy rate was significantly higher with 7.47 per 13
rate of 1.79 (þ/2 0.52 standard error) per 100 women after cycles (P , 0.00001) and 100 women (Table VII and
13 months of use (Table V); all unintended pregnancies due Figure 3). In 16 of the 22 pregnancy cycles, there was unpro-
to method and user failure were included. There was no differ- tected intercourse in the fertile time.
ence between the learning phase (first 3 months of use) and the
subsequent months of use. Discontinuation
We compared the rates of unintended pregnancies between Discontinuation for dissatisfaction or difficulties with the
the two groups ‘STM only’ and ‘STM mix’ (Table VI). From method, including change to another family planning method
was an important parameter of acceptability. The overall dis-
continuation rate for this category was 9.2 per 100 women at
Table IV. Socio-demographic characteristics of the cohort at study entry 13 cycles of method-use (Table VIII).
(n ¼ 900) The overall rate of lost to follow up was 6.7% after 13 cycles
Feature Categories %
(Table VIII).
Other reasons for discontinuation before cycle 13 given by
Age distribution (n ¼ 900) 19– 24 years 24.7 34% of the couples included: desire to achieve a pregnancy
25– 29 years 38.6
30– 35 years 24.7
(8%); separation from partner (2%); medical reasons (4%)
35– 39 years 8.9 and most frequently (22%) because they wished to discontinue
40– 45 years 3.2 participating in the study, although they wished to continue to
Highest educational level (n ¼ 891) Primary school 11.1
Secondary school 63.5
use the STM.
University degree 25.4
Occupation (n ¼ 880) Working or training 60.0 Study population duration of participation
Housewife 39.0
Unemployed 1.0
Figure 3 shows the study population over time. The study
Marital status (n ¼ 888) Married 35.8 started with 900 women, 322 of them using ‘STM only’ and
Unmarried 62.5 509 of them using ‘STM mix’. 69 women did not document
Divorced 1.7
Religion (n ¼ 885) Catholic 73.8
their sexual behaviour. Figure 4 shows that almost 70% of
Protestant 19.2 the couples participated in the study for at least 12 cycles.
Others 1.0 Less than 25% of the couples remained in the study for
None 6.1
No. of previous pregnancies (n ¼ 856) 0 51.9
longer than 24 cycles. Therefore, the time bias influencing
1– 2 34.7 the pregnancy rates according to the Trussell approach was
3 13.4 not too serious, especially taking into account that all the par-
Family planning intention (n ¼ 838) Spacer 57.4
Limiter 20.6
ticipants who joined the study were new STM users.
Undecided 22.0
Cycle range (n ¼ 900) up to 5 days 55.4
.5 days 44.6 Discussion
Where these numbers do not total 900, the reminder are women for whom We believe that this is a significant prospective cohort study
there is no information. of a clearly defined STM method that has several distinctive
1314
 Effectiveness of a fertility awareness-based method

Table V. Overall unintended pregnancy rates per 100 women according to the Kaplan– Meier approach for the whole cohort (n ¼ 900) cut at 24 cycles

Ordinal Women Cumulative Cumulative number Rate (SE) of

cycle exposed number of of unintended unintended
number cycles pregnancies pregnancies

1 900 900 0 0
3 846 2624 0 0
6 740 4945 4 0.52 (0.26)
9 618 6933 10 1.4 (0.44)
12 509 8571 11 1.57 (0.47)
13 434 9005 12 1.79 (0.42)
18 318 10 815 15 2.61 (0.7)
24 229 12 386 15 2.61 (0.7)

SE, standard error (Annotation: after 24 cycles we cut the analysis; i.e. seven unintended pregnancies after this time).

features that has ensured its quality, these include: a large
database; relatively low lost to follow-up rate; inclusion of
the teaching phase; documentation of all sexual behaviour
and classification of the pregnancies as intended or unintended
according to the intention before conception. To ensure the
certain method or do not wish to be randomized. In addition
with most family planning methods, it is impossible to blind
the couples from the allocated method unless comparing
certain different hormonal contraceptives or intrauterine
devices.

Downloaded from http://humrep.oxfordjournals.org/ by guest on June 17, 2013

quality of this study, we have described in depth the recruit-
ment, teaching methods and follow-up of the study partici-
pants, including the inclusion and exclusion criteria. In
addition, we have clearly defined both ‘pregnancy’ and ‘unin-
tended pregnancy’ as well as ‘perfect use’ and ‘imperfect use’
behaviour. Finally, we have described how we analysed the
data from this large prospective cohort study.
While we recognize that critics may argue that this study was
not a randomized controlled trial, it should be recognized that
very few FAB method studies are randomized controlled trials.
The majority are observational studies. This study was a
prospective cohort study which could be seen to be placed
from an evidence-based perspective, between the retrospective
case-control studies and the randomized clinical trials. The
only randomized clinical trials on methods of NFP (Wade
et al., 1981; Medina et al., 1980) are of limited use: they
showed huge recruitment problems and retention as well as
having a very strong selection bias (participants had to agree
to expect quite high failure rates while attracted by free
medical care at study entry), their results are therefore very
questionable (Grimes et al. 2004). Randomized controlled
trials are rarely used to investigate other family planning
methods either because most couples have a preference for a
A unique feature of this study was that we have applied
advanced analytical methods to our data. We used non-
parametric models to estimate pregnancy rates: we applied
the Kaplan – Meier survival curve to estimate the total preg-
nancy rate and the Trussell approach to calculate the pregnancy
rates according to the proper risk set. Nevertheless, we recog-
nize that these approaches are all influenced to some extent by
the differing risk of conception per cycle (Ecochard, 2006;
Dunson et al., 1999).
The analysis of the efficacy has demonstrated that a STM
that uses two indicators of fertility—temperature and cervical
secretion observations to determine the end of the fertile time
and cervical secretions plus a calculation to identify the onset
of the fertile time—is an effective and acceptable method of
family planning. The overall rates of unintended pregnancies
were 1.8% and the drop out rate for dissatisfaction with the
method was only 9.2 per 100 women after 13 cycles of
method use.
We have demonstrated that the STM is significantly more
effective to avoid pregnancies if used consistently and per-
fectly with couples abstaining from intercourse during the
fertile time: 0.4% pregnancy rate per year. We found similar
pregnancy rates for couples who occasionally use barrier

Table VI. Overall unintended pregnancy rates per 100 women according to the Kaplan– Meier approach within subcohorts ‘STM only’ (1, bold) and ‘STM
mix’ (2, italic) cut at 24 cycles

Ordinal cycle number Women exposed Cumulative number Rate of unintended pregnancies (SE)
of unplanned
pregnancies

1 2 1 2 1 2

1 322 509 0 0 0 0
3 295 489 0 0 0 0
6 248 448 2 2 0.75(0.53) 0.43(0.31)
9 202 381 4 6 1.62(0.89) 1.4(0.57)
12 165 324 4 7 1.62(0.89) 1.67(0.63)
13 136 280 4 8 1.62(0.89) 2.02(0.72)
18 102 205 6 9 3.33(1.44) 2.45(0.83)
24 66 154 6 9 3.33(1.44) 2.45(0.83)

1315
P.Frank-Herrmann et al.

Downloaded from http://humrep.oxfordjournals.org/ by guest on June 17, 2013

Figure 2. Comparison of the rates of unintended pregnancies per 100 women and their standard error between the group ‘STM only’ (B) and the
group ‘STM mix’ (O) at different cycle numbers using the Kaplan– Meier analysis.

methods, mainly condoms, during the fertile time as compared
to couples who were abstinent. However, there were very few
pregnancies in the two samples. We therefore barely had
enough statistical power to evaluate the multivariate and
adjusted effect of barrier methods for avoiding pregnancy.
The use-effectiveness rates (¼ total unintended pregnancy
rates) compare very well with the results of the European
study and with the interim results of the German database, and
with the symptothermal subgroup of the Italian study
(Frank-Herrmann et al., 1991; Barbato and Bertolotti, 1988;
The European Natural Family Planning Study Groups, 1999).
However, the two latter studies were still using the Pearl
formula as statistical method. The previous International Rice
Fairfield study, that used a NFP-method that preceded the
STM, showed reasonable results for developed countries (Rice
et al., 1981). The markedly high use-effectiveness rates of our
data may partly be explained by the motivation of those
couples and their teachers who agreed to participate in the study.
When comparing different methods of family planning,
method effectiveness rates are more frequently quoted than
the use-effectiveness rates which are strongly dependent on
the selection of the study population. The fact that variations
in study population make it more difficult to interpret the
overall effectiveness of the FAB methods has been discussed
in depth (Kambic, 1999).
For a contraceptive method to be rated highly efficient as the
hormonal pill, it requires a method failure rate of less than one
pregnancy per 100 women per year. Our method-effectiveness
of 0.4% can be interpreted as one pregnancy occurring per
3250 cycles (assuming a 13 cycle year). We therefore maintain
that the method effectiveness of the STM investigated in this
study is comparable to the method effectiveness of modern
contraceptive methods like oral contraceptives.
The authors were surprised by the high efficacy during
additional barrier method use. We did not find any differences
in pregnancy rates between STM only users and STM mix
users. Obviously, couples with fertility awareness knowledge
are more likely to use condoms more consistently in the
fertile time. Most cited NFP studies do not report the quantity
of additional barrier method use, yet we have learnt from the
European study that it exists to a certain extent within all com-
munities that use NFP methods.

Table VII. Rates of unintended pregnancies per 100 women and year according to sexual behaviour during the fertile time

Cycles Unintended pregnancies

n % n % per yeara 95% CI lower limitb 95% CI upper limitb

Abstinence in the fertile time 6022 34.14 2 0.43 0.05 1.55

Protected intercourse in the fertile time 4375 24.80 2 0.59 0.07 2.13
Unprotected intercourse in the fertile time 2353 13.34 14 7.46 4.15 1.23
Unprotected and protected intercourse in the fertile time 1183 6.71 2 2.18 0.27 7.65
Genital contact or coitus interruptus in the fertile time 1080 6.12 1 1.20 0.03 6.50
No documented sexual behaviour 2625 14.88 1 0.49 0.01 2.72
Total 17 638 100 22 1.61 1.01 2.43
a
Rate per 13 cycles and 100 women, calculated according to the formula 100  (1 2 (1 2 P)13), P ¼ probability per cycle (not displayed).
b
CI ¼ confidence interval, calculated out of the confidence intervals of the pregnancy rates per cycle (not displayed).

1316

Figure 3. Rates of unintended pregnancies per 100 women and year with 95% confidence intervals in relation to sexual behaviour during the
fertile time according to a modified ‘perfect/imperfect-use’ approach.
For the risk taking couples who had unprotected intercourse
during the fertile time, the pregnancy rate increases up to 7.5%
per year. We acknowledge that this is surprisingly low.
However, one has to realize that the median fertile time deter-
mined by the STM is 13 days a cycle (less days after the first
year). The potential fertile time is in fact longer than the
actual physiological fertile time. Therefore, we recognize that
some of the couples were practising conscious intelligent risk
taking, i.e. no unprotected intercourse during the few highly
fertile days and intercourse only occurred on days at the
margins of the beginning and end of the fertile time that
would be considered to be a relatively low fertile time. The
pregnancy rates during the different days of the identified
fertile time varied a lot according to their interval between
the day of sexual intercourse and their distance to the estimated
day of ovulation (Gnoth et al., 2003). To summarize, one
cannot compare the pregnancy rate on possibly fertile days—
derived from couples who explicitly wanted to avoid a preg-
nancy and therefore practised ‘intelligent risk taking’—with
pregnancy rates derived from intention to get pregnant.
In addition, we acknowledge that the pregnancy rates during
the teaching phase which included the first three cycles were
shown to be as low as in the subsequent cycles. We suggest
this may be attributed to the high quality training and supervi-
sion of the local STM teachers as well as the standardized
teaching materials (Arbeitsgruppe NFP, 2004).
Our findings suggest that compared to other FAB methods,
e.g. the Billings method, the Creighton model or other cervical
secretion methods, a method such as the German STM that uses
two indicators of fertility – cervical secretions plus a calculation
to identify the onset of the fertile time and basal body tempera-
ture and cervical secretions observation to determine the end of
the fertile time – is an effective and acceptable method of
family planning (World Health Organization, 1981a,b and
1983). In the last two decades, several effectiveness studies
Effectiveness of a fertility awareness-based method
Downloaded from http://humrep.oxfordjournals.org/ by guest on June 17, 2013
have been carried out in developing countries (Thapa et al.,
1990, Xu et al., 1994; Arevalo et al., 2004). We believe that
the social setting and infrastructure to deliver the FAB
methods in these countries is very different from that of the
European countries and it is difficult to compare our results
directly with these groups. We have therefore only considered
studies carried out within the last 25 years within the developed
world to compare our results with. There is only a small
number of European effectiveness studies based on cervical
secretion as a single indicator method, due to the fact that
these single indicator methods are not used very frequently in
Europe. It is interesting to note that in the WHO five-country
study, the pregnancy rates of the ‘Billings method’ (cervical
secretion ¼ single indicator) was much higher in the two
industrial countries compared to the developing countries: in
Ireland with 5.1 pregnancies per 100 women years and 9.4 in
New Zealand (World Health Organization 1981a,b).
Pregnancy rates can also be biased if the studies include
participants who are likely to have very low fertility, for
example, if they include participants who are fully breast-
feeding, or they can be biased downward if they include
Table VIII. Dropout rates with their SE according to the Kaplan– Meier
analysis, cut at 24 cycles (n ¼ 900 women)
Ordinal cycle Rate of dropout for Rate of lost to
number dissatisfaction follow-up
1 0.22(0.16) 0.22(0.31)
3 1.02(0.34) 0.80(0.59)
6 2.4(0.53) 2.47(1.07)
9 4.39(0.74) 3.95(1.39)
12 8.78(1.11) 6.30(1.86)
13 9.20(1.15) 6.73(1.94)
18 12.18(1.4) 8.57(2.33)
24 16.12(1.74) 10.87(2.92)
1317
P.Frank-Herrmann et al.
Figure 4. Quantitative development of the study population over time (the cohorts ‘only’ and ‘mix’ do not sum up to ‘overall’ because of 69
participants who did not report their sexual behaviour, see above).
those cycles from women who are trying to achieve a preg-
nancy, while excluding all the planned pregnancies from the
unintended pregnancy data. (Hilgers and Stanford, 1998;
Howard and Stanford, 1999). The Italian STM study had two
groups of participants; one group used two parameters to deter-
mine the onset of the fertile time and the other group only
observation of cervical secretions. The study found that all of
the 12 method-related unintended pregnancies occurred in
the group that used only cervical secretion rules to determine
the onset of the fertile time (460 women, 8140 cycles).
Studies that have investigated the efficacy of the hormonal kit
Persona and the Computer thermometers have found that the
effectiveness is not as high as double-check variation of the
STM (Freundl et al., 2003).
Conclusions
This is the first time that a large STM database has been estab-
lished with sufficient detailed information on sexual behaviour.
It enables the true method effectiveness for the STM to be cal-
culated. Our results show that 0.4 unintended pregnancies
occurred per 100 women years, if there was abstinence
during the fertile time. In addition, our results showed that
when barrier methods were used during the fertile time the
rate of unintended pregnancies was not significantly different.
The use-effectiveness of the method, i.e. the overall pregnancy
rate was 1.8% after 13 cycles of use and the discontinuation
rate due to dissatisfaction with the STM was only 9.2 per
100 women after 13 cycles; this demonstrates a fairly good
acceptability of a FAB method that uses two indicators of fer-
tility – cervical secretions plus a calculation to identify the
onset of the fertile time and temperature and cervical secretion
observations to determine the end of the fertile time.
1318
Downloaded from http://humrep.oxfordjournals.org/ by guest on June 17, 2013
Acknowledgements
The authors cordially thank Silvia Heil-Schlehuber, NFP study centre,
for her assistance in the follow-up procedures and her valuable
comments.
References
Arbeitsgruppe NFP (2006) Natürlich und sicher. 8th edn. Trias, Stuttgart.
Arevalo M, Jennings V, Nikula M and Sinai I (2004) Efficacy of the new
TwoDay Method of family planning. Fertil Steril 82,885– 892
Barbato M and Bertolotti G (1988) Natural Methods for fertility control: a
prospective study. Int J Fertil Suppl, 48–51.
De Leizaola-Cordonnier A (1995) Natural family planning effectiveness in
Belgium. Adv Contracept 11,165–172.
Dunson DB, Baird DD, Wilcox AJ and Weinberg CR (1999) Day-specific
probabilities of clinical pregnancy based on two studies with imperfect
measures of ovulation. Hum Reprod 14,1835–1839.
Ecochard R (2006) Heterogenity in fecundability studies: issues and modelling.
Stat Methods Med Res 15,141–160.
Fehring RJ, Lawrence D and Philpot C (1994) Use effectiveness of the
Creighton model ovulation method of natural family planning. J Obstet
Gynecol Neonatal Nurs 23,303–309.
Frank-Herrmann P, Freundl G, Baur S, Bremme M, Döring GK, Godehardt E
and Sottong U (1991) Effectiveness and acceptability of the symptothermal
method of natural family planning in Germany. Am J Obstet Gynecol
165,2052– 2054.
Freundl G, Godehardt E, Kern PA, Frank-Herrmann P, Koubenec HJ and
Gnoth Ch (2003) Estimated maximum failure rates of cycle monitors
using daily conception probabilities in the menstrual cycle. Hum Reprod
18,2628 – 2633.
Gnoth C, Bremme M, Klemm R, Frank-Herrmann P, Godehardt E and
Freundl G (1999) Research and quality control in natural family
planning with relational database systems. Adv Contracept 15,375 – 380.
Gnoth C, Frank-Herrmann P, Schmoll A, Godehardt E and Freundl G (2002)
Cycle characteristics after discontinuation of oral contraceptives. Gynecol
Endocrinol 16,307– 317.
Gnoth C, Godehardt D, Godehardt E, Frank-Herrmann P and Freundl G (2003)
Time to pregnancy: results of the German prospective study and impact on
the management of infertility. Hum Reprod 18,1959– 1966.
Gomes I and Congdon D (1988) The natural family planning programme in
Bangladesh. Int J Fertil 33,52–53.

Grimes DA, Gallo MF, Grigorieva V, Nanda K and Schulz KF (2004) Fertility
awareness-based methods for contraception. Cochrane Database Syst Rev
18(4), pCD004860.
Hilgers TW and Stanford JB (1998) Creighton Model NaPro Education
Technology for avoiding pregnancy: use effectiveness. J Reprod Med
43,495– 502.
Howard MP and Stanford JB (1999) Pregnancy probabilities during use
of the Creighton model fertility care system. Arch Fam Med 8,391–402.
Indian Council of Medical Resarch Task Force on Natural Family planning
(1996) Field trial of billings ovulation method of natural family planning.
Contracept 53,69– 74.
Jennings V and Sinai I (2001) Further analysis of the theoretical effectiveness
of the two day method of family planning. Contracept 64,149–153.
Kambic RT (1999) The effectiveness of natural family planning methods
for birth spacing: a comprehensive review In Girotto S and Bressan F
(eds.) Human fertility Regulation – Demographic and Statistical Aspects.
Edizioni libreria cortina, Verona, pp. 63– 90.
Kaplan EL and Meier P (1958) Nonparametric estimation from incomplete
observations. J Am Stat Ass 53,457–481.
Matthews DE and Farewell VT (1996) Using and understanding medical
statistics. Karger, Basel-Freiburg-Paris.
Medina JE, Cifuentes A, Abernathy JR, Spieler JM and Wade ME (1980)
Comparative evaluation of two methods of natural family planning in
Colombia. Am J Obstet Gynecol 138,1142– 1147.
Potter RG (1966) Application of life-table techniques to measurement of
contraceptive effectiveness. Deomgraphy 3,297–304.
Potter LS (1996) How effective are contraceptives? The determination and
measurement of pregnancy rates. Obstet Gyncecol 88,13– 23.
Raith E, Frank P and Freundl G (1999) Natürliche Familienplanung heute – mit
ausführlicher Darstellung der Zykluscomputer, 3rd edn. Springer, Berlin,
Heidelberg, New York.
Rice FJ, Lanctot CA and Garcia-Devesa C (1981) Effectiveness of the
sympto-thermal method of natural family planning: an international study.
Int J Fertil 26,222–230.
Effectiveness of a fertility awareness-based method
Thapa S, Wonga MV, Lampe PG, Pietojo H and Soejoenoes A (1990) Efficacy
of three variations of periodic abstinence for family planning in Indonesia.
Stud Fam Plann 21,327– 334.
The European Natural Family Planning Study Groups (1999) European
multicenter study of natural family planning: efficacy and drop-out. Adv
Contracept 15,69–83.
Tietze C and Lewit S (1974) Statistical evaluation of contraceptive methods.
Clin Obstet Gynecol 17,121–138.
Trussell J and Grummer-Strawn L (1990) Contraceptive failure of the ovulation
method of periodic abstinence. Fam Plann Perspect 22,65– 75.
Trussell J and Grummer-Strawn L (1991) Further analysis of
contraceptive failure of the ovulation method. Am J Obstet Gynecol
165,2054–2060.
Trussel J and Kost K (1987) Contraceptive failure in the United States: a critical
review of the literature. Stud Fam Plan 18,237–283.
Wade ME, Mc Carthy P, Braunstein GD, Abernathy JR, Suchindran CM and
Harris GS (1981) A randomized prospective study of the use-effectiveness
of two methods of natural family planning. Am J Obstet Gynecol
141,368– 376.
World Health Organization (1981a) A prospective multicentre trial of the
ovulation method of natural family planning. I. The teaching phase. Fertil
Steril 36,152– 158.
World Health Organization (1981b) A prospective multicentre trial of the
ovulation method of natural family planning. II. The effectiveness phase.
Downloaded from http://humrep.oxfordjournals.org/ by guest on June 17, 2013
Fertil Steril 36,591–598.
World Health Organization (1983) A prospective multicentre trial of the
ovulation method of natural family planning. III. Characteristics of the
menstrual cycle and of the fertile time. Fertil Steril 40,773–778.
Xu JX, Yan JH, Fan DZ and Zhang DW (1994) Billings natural family planning
in Shanghai China. Adv Contracept 10,195–204.
Submitted on September 4, 2006; resubmitted on November 8, 2006;
resubmitted on December 7, 2006; accepted on January 2, 2007
1319